Add new module - Introduction to green computing #414
Description
Title
Introduction to green computing
Topic keywords
computing resources, environmental impact, neuroimaging
Module Description
Neuroimaging is an energy-demanding field - from data collection to machine learning analyses to large-scale data storage. This module introduces strategies to evaluate and implement more sustainable neuroimaging pipelines, and encourages trainees to consider the environmental aspects related to scientific activities. The environmental impact of AI will be discussed, covering both its usage as a tool (e.g., generative assistant), and its NeuroAI specific applications (i.e., data analysis).
Key tools / technology included in this module
- Online carbon footprint calculators
- Python-based tools (e.g., cats, codecarbon)
Prerequisites
- Installation
- Introduction to python
- HPC (optional)
Study outcomes
- Identify energy-intensive stages of neuroimaging workflows
- Apply carbon tracking tools (measure & report energy usage / carbon footprint)
- Apply green computing principles to neuroimaging workflows (strategies to reduce)
Estimated study time
~2h-3h
Exercise examples
- Calculate environmental footprint based on log information out of slurm scheduler (files provided)
- Provide an article and ask the student to estimate the energy usage and carbon intensity using an online calculator given the information regarding the computing resources reported.
- Use a carbon tracking tool to measure the energy usage for a given task (analysis).
- Ask students to modify different aspects of the task to see how that could affect the energy usage.
References you would like to include
No response
Things to check by the reviewers.
- Use free, open source tools.
- Python-based tool.
- Be of high quality.
- Have a broad appeal.